2026-09-07 AI Agent Identity News
Summary: Second digest in the series. International commentary elaborates what an Estonian AI-agent registration framework might look like, while the US, OWASP, and IETF-adjacent standards work advance agent inventories, runtime control, and the separation of discovery from identity.
Sources: 2026-09-07-ai-agent-identity-news.md
Last updated: 2026-09-08
Note on one item below: this report’s Estonia section describes a registration / ID-code framing that was materially corrected the following day. See 2026-09-08-ai-agent-identity-news and the Correction section of estonia-ai-agent-id before relying on it.
Estonia and Estonia-related coverage
TechRadar Pro published secondary commentary by an e-Residency leader outlining what an Estonia-based registration framework could look like: a registered numeric identifier linked to one or more agents, bounded authorizations, a human identity anchor who remains liable, and straightforward revocation when an agent’s mandate ends (source: 2026-09-07-ai-agent-identity-news.md, citing TechRadar Pro, published 2026-09-04).
The report notes this adds delegation, liability, revocation, and auditability detail to the June announcement covered in 2026-09-04-ai-agent-identity-news. The delegation, liability and revocation elements survive the later correction; the “identifier issued to the agent” framing does not — see estonia-ai-agent-id.
No new Estonia-origin source matching the full literal keyword pattern tehisaru + isikukood + (agendid OR assistendid OR aruait) was found in this run. The literal term aruait again produced no match.
Developments in other countries
United States — proposed Stop Rogue AI Act. A bill introduced by Representatives Josh Gottheimer and Mike Lawler would direct NIST to develop standards for secure AI-agent deployment: continuous verification of agent actions, security and reliability evaluation, tamper-proof logs, and a machine-readable inventory of deployed agents (source: 2026-09-07-ai-agent-identity-news.md, citing Axios, published 2026-09-03). See united-states-ai-agent-policy.
Standards work
OWASP Agent Control Standard. A primary standards project defining an open foundation for inspectable, traceable, instrumentable agents, with runtime enforcement of declarative safety policies through portable middleware hooks (source: 2026-09-07-ai-agent-identity-news.md, citing OWASP, published 2026-09-01).
IETF-related analysis separating discovery from identity. The Cloud Security Alliance notes the proposed IETF DAWN work focuses on discovering already-identified agents, while identity and authorization work develops separately around WIMSE, OAuth delegation, and SPIFFE-style workload identifiers (source: 2026-09-07-ai-agent-identity-news.md, citing Cloud Security Alliance, published 2026-09-02).
Both items are covered on agent-standards-and-interoperability.
Research
2026-09-07 was a Monday. Rather than a separate research section, the source’s frontmatter states its coverage window “also includes newly surfaced relevant papers and standards work” — the OWASP and CSA items above are that material. No distinct academic papers or preprints were listed (source: 2026-09-07-ai-agent-identity-news.md).